Ctrl+Enter to process
9
Autoregressive Generation
LLMs generate text one token at a time, in a loop. After picking a token, that token is added to the input and the whole process runs again from the top to pick the next token. It's like writing a sentence one word at a time, where each new word depends on everything you've written so far. This loop continues until the model produces a special "stop" token or hits a length limit.
Common misconception: The model does not plan ahead or outline the full response before writing it. It generates each token purely based on what came before. There is no "draft then revise" step — what gets generated first stays in the context and influences everything that follows.
Enter a query above to begin.